Albumin Excretion Rate and Cardiovascular Risk Could the Association Be Explained by Early Microvascular Dysfunction? W. David Strain,1 Nish Chaturvedi,1 Christopher J. Bulpitt,2 Chakravarthi Rajkumar,2 and Angela C. Shore3
Elevated albumin excretion rate (AER) independently predicts total and cardiovascular mortality in a variety of conditions, although the exact mechanisms are unknown. Laser Doppler fluximetry was used to study associations with risk factors and renal damage (AER calculated from a timed overnight urine collection) in 188 people without diabetes and 117 individuals with diabetes. Skin flow (flux) in response to arterial occlusion (ischemia) was measured. Three distinct patterns of postischemic peak flow were observed: 1) gradual rise to peak (normal), 2) nondominant early peak, and 3) dominant early peak. Those with a dominant early peak were more likely to have diabetes (P ⴝ 0.01), hypertension (P ⴝ 0.001), and obesity (P < 0.001) and had a higher AER (12.6 g/min [95% CI 7.8 –20.2] vs. 7.2 [5.5–9.5] nondominant early peak group and 3.7 [3.2– 4.1] normal group; P < 0.001 for trend). This could not be accounted for by conventional cardiovascular risk factors (P < 0.001 after adjustment). A rapid peak flow response after ischemia is associated with an elevated AER and increased cardiovascular risk. This may represent shared mechanistic pathways and causative or consequential changes in the microvasculature and supports the hypothesis that microvascular dysfunction may contribute to large vessel pathophysiology. Diabetes 54: 1816 –1822, 2005
E
levated albumin excretion rate (AER) is a hallmark of microvascular damage in those with diabetes and independently predicts total and cardiovascular mortality (1). Similar observations in hypertensive patients (2), postmenopausal women (3), elderly patients (4), and the general (largely nondiabetic) population, right into the normal range of AER From the 1International Centre for Circulatory Health, Faculty of Medicine, Imperial College London at St Mary’s, London, U.K.; 2Care of the Elderly, Faculty of Medicine, Imperial College London at the Hammersmith Hospital, London, U.K.; and 3Clinical Microvascular Research, Institute of Biomedical and Clinical Science, Peninsula Medical School, Exeter, U.K. Address correspondence and reprint requests to Professor Angela Shore, Clinical Microvascular Research, Institute of Biomedical and Clinical Science, Peninsula Medical School, Barrack Rd., Exeter EX2 5AX, U.K. E-mail: angela.
[email protected]. Received for publication 27 August 2004 and accepted in revised form 21 February 2005. AER, albumin excretion rate. © 2005 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1816
(5–7), have led to speculation regarding the link between AER and the etiopathogenesis of cardiovascular disease. The relationship is unlikely to be causal. AER may be a marker of unmeasured risk factors or an index of microvessel damage in other vascular beds, which in turn either results in or is associated with macrovascular disease. This association is supported by the observation that microvascular damage in other organ beds, such as the eye, is highly predictive of future cerebrovascular disease independent of conventional risk factors (8). Furthermore, in a population of healthy women, microvascular responsiveness was inversely related to Framingham risk score (9), suggesting that the link between microvascular dysfunction and cardiovascular risk may span into the healthy population. Therefore we investigated the hypothesis that abnormalities in AER reflect abnormalities of microvascular function that are mirrored in other vascular beds, by examining the patterns of skin microvascular perfusion after an ischemic stimulus in a population with AERs ranging from normal to elevated. Potential explanations for any links observed were also explored. RESEARCH DESIGN AND METHODS This was a secondary analysis of data designed to compare microvascular function in people with and without diabetes. Men and women from a family practitioner register, aged 40 – 65 years and of African Caribbean and European descent, were randomly sampled and stratified by age-group, sex, and ethnicity to yield 188 nondiabetic and an additional 117 participants with type 2 diabetes. The family practitioner register provides a comprehensive sampling frame, because more than 97% of the population is registered (10), and notification of diabetes status is a requirement under the family practitioner contract and the English National Service Framework (http://www.doh.gov. uk/nsf/diabetes, accessed 21 August 2003). The following groups of people were excluded: those with cancer or major psychiatric illness, which would have precluded travel and attendance for investigation; those with known HIV or who were hepatitis B positive, because of the infection hazard posed as a consequence of blood sampling; and those with sickle cell disease or atrial fibrillation, which would interfere with the vascular tests. Respondents attended the Hammersmith Hospital for examination. The protocol was approved by the medical research ethics committee of the Hammersmith Hospital. All participants gave informed written consent and completed a questionnaire on demographics, medical history, medications, and lifestyle behaviors. Ethnicity was self-assigned and cross-checked against patients’ own and parental country of birth. Measures to a standard protocol included height, weight, and waist-to-hip ratio (11). Blood and urine samples were analyzed by the on-site laboratory, which is a member of the appropriate U.K. National Quality Assessment Scheme. Glucose and lipid levels were determined using the AU600 Olympus Diagnostic analyzer, and insulin was determined by the Abbott Axsym immunometric assay. All participants attended in the morning, after fasting from at least midnight the night before and having refrained from smoking, drinking tea or coffee, and taking any medications. The mean of DIABETES, VOL. 54, JUNE 2005
W.D. STRAIN AND ASSOCIATES
FIG. 1. Distribution of time to peak responses. three supine resting brachial and ankle blood pressure readings using an automated device (Omron 705CP) was taken as the representative value of brachial and ankle blood pressures, from which ankle-brachial indexes were calculated. Pulse wave velocity, a measure of arterial stiffness, was measured in the arterial segment between the femoral and the dorsalis pedis pulse to assess arterial function within the supply vessels to the foot using the Complior device (COLSON, Paris, France) according to a standard protocol (12). This was performed to explore whether abnormalities at the microvascular level were associated with macrovascular dysfunction or localized to the microcirculation. Microvascular assessment was performed using laser Doppler fluximetry (DRT4; Moor Instruments, Axminster, U.K.; time constant, 0.1 s; bandwidth, 22 kHz) with the probe sited on the dorsum of the foot. Subjects were allowed to acclimatize for a minimum of 40 min lying supine in a quiet temperaturecontrolled room (at 22°C). A 3-min period of ischemia was induced by inflation of a sphygmomanometer cuff around the calf to a suprasystolic level. Adequate occlusion was determined as the loss of pulsatile flux on visual inspection of the trace, followed by decline of flow to a steady constant level (the biological zero). Rapid release of cuff pressure prompted a reactive increase in flux (flow) to an eventual peak, which was recorded digitally using the DRT4 For Windows Version 1.2 software (Moor Instruments) sampling at 40 Hz. Before release of the cuff, subjects were instructed to remain as still as possible while the cuff was released and for the following 2 min. However, if involuntary movement was observed, the tracing was discarded. The tracing was not repeated because it has been suggested that a second arterial occlusion within a short time frame may result in a different response. Flux and the time to reach the peak flux were recorded. A timed overnight urine sample was collected, and albumin and creatinine concentrations were measured using the immunoturbometric method, with a lower level of sensitivity of 5 g/dl. All values below this were coded as 5 g/dl. AER and albumin-to-creatinine ratio were calculated according to standard formulas. Analysis. Skewed variables were log transformed. Statistical significance for categorical variables was calculated using the 2 test for trend and the Student’s t test for continuous variables. ANCOVA was used to determine explanations for differences in key variables between the three peak response curve morphologies, with tests for interaction. In keeping with the recommendations of Cupples et al. (13) and Rothman (14), the measured significance of the variables of interest is reported without adjustment for multiple testing. Bonferroni’s adjustment is only relevant for additional covariates within the models, and where significance of these variables is described, this is after adjustment. Analyses were performed using STATA (version 7). Analyzable microvascular recordings were available for 271 of 305 participants. The rest were discarded because of poor quality either because of patient factors or technical difficulties (29 because of movement artifact identified at the time of recording, 4 because of lost data, and 1 because of DIABETES, VOL. 54, JUNE 2005
inability to occlude the artery). The demographics of the 34 patients with discarded data did not differ from those of the remaining 271 patients. There was no ethnic difference in the distribution of the peak response curve morphologies or in the correlates thereof; thus the ethnic groups were combined. A frequency distribution plot of times to peak flux postischemia indicated the presence of at least two distinct subgroups (Fig. 1). A detailed visual inspection of each trace, masked to all patient data apart from randomly allocated study number, was then performed. The principal criteria used were the presence of a transient peak in the first 2 s and whether this peak or a subsequent later peak was the tallest (Fig. 2). This analysis revealed the presence of a group of individuals with a “normal” graded rise to peak response curve (with a peak at 9.7 s [95% CI 8.5–11.1]) and a group with a dominant early peak within 2 s, which declined rapidly and was then followed by a lesser “normal” peak at 11.9 s (9.7–14.0). An “intermediate” group, those also with an early peak within 2 s but which was lower than the subsequent “normal” graded rise to peak at 11.7 s (10.1–13.5), was also identified. These three groups were labeled according to their reactive hyperemia peak response curve morphology, as “group 1-normal peak response curve morphology,” “group 2-nondominant early peak,” and “group 3-dominant early peak” (Fig. 3). The intraobserver repeatability of this visual analysis and categorization was 100% (n ⫽ 25; time separation, 3 months). In an additional 22 individuals studied on two separate occasions (10 with proven cardiovascular disease and 12 normal control subjects studied by two separate investigators) at least 24 h apart, the within-subject repeatability of category membership was also high ( ⫽ 0.90; P ⬍ 0.0001). The investigators were blinded to the results of the previous analysis when these repeatability studies were performed.
RESULTS
Cardiovascular disease risk factor status varied by the microvascular response to ischemia. There was a gradation in the severity of cardiovascular risk factors across the microvascular response groups (Table 1), such that those individuals with a dominant early peak (group 3) had higher blood pressures and glucose levels, had greater obesity, and were more likely to be ex-smokers or current smokers than those with normal peak response curve morphology (group 1). In a multivariate model, independent associates of trend across the three peak response curve morphology groups were mean resting systolic blood pressure (-regression coefficient [95% CI] across the three groups 0.0058 [0.0005– 0.0110]; P ⫽ 0.03), noctur1817
MICROVASCULAR REACTIVITY AND AER
FIG. 2. Algorithm for the assessment of time to peak tracings. *Transient peak: a peak that has reached its zenith within the first 2 s and has a total duration of
